can tf.data.experimental.CsvDataset work with tab as a field sep?

notes.txt

It looks like copy and paste of the tsv file replaced tabs with spaces.  Here is where the original file can be found.

https://waterdata.usgs.gov/nwis/dv?cb_00045=on&cb_62620=on&format=rdb&site_no=0204288721&referred_module=sw&period=&begin_date=2018-04-12&end_date=2019-04-12

testy.py

from __future__ import absolute_import, division, print_function

from pathlib import Path

import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns


import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers

print(tf.__version__)

# assumes .tsv file is in a directory named "tsv" above the source file
file_path = Path('../tsv')
file_name = 'SN0204288721.tsv'
file_path = file_path.joinpath(file_name)

# skip the comments but not the headers
# the comments are at lines 1-31
# the column headings is at line 32
# the 5s 15s 20d 14n ... line ends is at 33
num_lines_to_skip = 33

# use csv file reader but with tab as separator
text_lines_ds = tf.data.experimental.CsvDataset(filenames=str(file_path),
                                                record_defaults = [tf.string,  # usgs
                                                                   tf.int64,   # site_no
                                                                   tf.string,  # date_time
                                                                   tf.float32,  # Tidal High
                                                                   tf.string,  # Tidal High qual code
                                                                   tf.float32,  # Tidal Low
                                                                   tf.string,  # Tidal Low qual code
                                                                   tf.float32,  # Precipitation (inches)
                                                                   tf.string], # Precip qual code
                                                header=False,
                                                field_delim='\t' # trying to use tab sep
                                               ).skip(num_lines_to_skip)